Guide
AI training data with EU data residency
Residency is not a checkbox on a vendor's website. It is four questions with written answers: where the data is stored, where it is processed, who can see it, and which company can be compelled to hand it over.
Buyers in healthcare, finance, the public sector and anything under the AI Act ask the same question in different words: can the data stay in Europe. For model training data the answer is harder than for a SaaS tool, because the data is looked at by people. Where those people sit is part of residency.
This guide lists what to require, where vendors typically fall short, and how Pathwize answers each point.
questions that decide residency
places vendors usually fail it
of Pathwize storage, processing and experts in the EU
downloads needed for sensitive data
Vendor facts come from the vendors' own material and widely cited public sources as of October 2026; pricing is per quote unless the vendor publishes it. Pathwize is EU-native: all data stays within the EU.
What to require in writing
01
Storage and processing in the EU
Named regions and providers for every system that touches the data, including the labeling tool, the file store and backups.
02
Annotators in the EU
The countries of the people who see the data, and a way to restrict access to the EU or to specific countries. Global crowds fail this by default.
03
An EU contracting party without a compellable parent
A US parent company can be required to produce data under the CLOUD Act regardless of where the EU subsidiary stores it.
04
Sub-processor list and change notice
Cloud, support tooling, model providers used for pre-labeling. Each one outside the EU is a transfer.
05
No-download handling for sensitive data
Annotators work in the browser; nothing lands on a laptop. Ask whether the vendor can do this.
06
Deletion and audit
Deletion on request with confirmation, and an audit trail of who accessed what and when.
How Pathwize answers each point
Entity: Pathwize is operated by Note Buddy's GmbH, Cologne, Germany (Amtsgericht Bonn, HRB 25657). No non-EU parent company.
Infrastructure: storage and processing on EU regions; the labeling workspace runs in the browser, items are served from EU storage with short-lived links, and sensitive tasks can be configured without download.
People: experts from the 200k+ network live in Europe and work under clean contractor classification. Access can be limited to named countries per task.
Evidence: each label carries the expert, the timestamp and the time spent; AI interviews are recorded and integrity-checked; accounts and their data can be deleted on request, with the deletion confirmed.
Exports in CSV, JSON or COCO, an API and webhooks, so the data moves into your pipeline without a third tool in between.
FAQs
The ones that put residency in the processing agreement and can show EU storage, EU processing and EU-based annotators. Pathwize does; for other providers see our comparison of domain expert data providers in Europe.
Legally, Switzerland holds an EU adequacy decision, so transfers are permitted under GDPR. Whether your policy accepts Switzerland is your call; many Swiss buyers in turn require data to stay in Switzerland, which Pathwize can arrange per task.
Article 10 sets data governance duties for high-risk systems: documented provenance, relevance, representativeness and error handling. Provenance per item, which Pathwize records by default, is the part most vendors cannot show.
Yes. Pools can be limited by country, and the German-speaking network is large enough for most tasks. The German buyer pages describe this in detail.
Usually within a few days. You send the task, the classes or rubric and a few examples; qualified experts take the test on your data; you approve who continues. A 50-item pilot is free.
Per hour of expert work, quoted per task depending on the domain and the complexity. Time is tracked automatically per item, so you see cost per item and per class in the dashboard and in the export.
In the EU. Pathwize is based in Cologne, Germany; storage and processing run on EU infrastructure, and experts work from Europe. Sensitive data can be worked in a no-download environment.
Related